2082 Biophysical Journal Volume 107 November 2014 2082–2090 Article
Physical Properties of Escherichia coli Spheroplast Membranes
Yen Sun,1 Tzu-Lin Sun,1 and Huey W. Huang1,* 1Department of Physics & Astronomy, Rice University, Houston, Texas
ABSTRACT We investigated the physical properties of bacterial cytoplasmic membranes by applying the method of micropi- pette aspiration to Escherichia coli spheroplasts. We found that the properties of spheroplast membranes are significantly different from that of laboratory-prepared lipid vesicles or that of previously investigated animal cells. The spheroplasts can adjust their internal osmolality by increasing their volumes more than three times upon osmotic downshift. Until the spheroplasts are swollen to their volume limit, their membranes are tensionless. At constant external osmolality, aspiration increases the surface area of the membrane and creates tension. What distinguishes spheroplast membranes from lipid bilayers is that the area change of a spheroplast membrane by tension is a relaxation process. No such time dependence is observed in lipid bilayers. The equilibrium tension-area relation is reversible. The apparent area stretching moduli are several times smaller than that of stretching a lipid bilayer. We conclude that spheroplasts maintain a minimum surface area without tension by a mem- brane reservoir that removes the excessive membranes from the minimum surface area. Volume expansion eventually exhausts the membrane reservoir; then the membrane behaves like a lipid bilayer with a comparable stretching modulus. Interestingly, the membranes cease to refold when spheroplasts lost viability, implying that the membrane reservoir is metabolically maintained.
INTRODUCTION
Direct probing of the physical properties of cell membranes (23), and daptomycin (24). To understand the in situ activ- has been performed on various animal cells with the ities of membrane-acting antibiotics, we need to know the methods of micropipette aspiration (1–5) and tether pulling physical properties of the target membranes (25). (6–10). These physical studies have provided invaluable As inhabitants of natural environments, bacteria have insight into the mechanical properties of eukaryotic cell abilities to survive very large changes of external osmolality membranes and thus a better understanding of the mem- (17). In E. coli, the cytoplasm responses to the changes of brane’s functions in cell biology (1,8,9,11). In comparison, external osmolality by adjusting its solute content and the the cytoplasmic membranes of bacteria are much less ac- amount of water (17,26–30). As a result, the cytoplasmic os- cessible to experimental study because they are normally motic pressure could exceed that of the surrounding medium shielded by outer membranes. Experimentalists have made by 0.5 to 3 atm, as the external osmolality decreases from use of spheroplasts, the cells from which the outer mem- 0.5 to 0.03 osmole/kg (Osm) (30). The cytoplasmic mem- branes have been removed, for patch-clamp, fusion, and brane cannot sustain such a large pressure drop. Cayley other experiments (12–14) and also for antibiotic studies et al. (30) have showed that the periplasmic solutes and vol- (15). However the physical properties of spheroplast mem- ume also changed with external osmolality, and concluded branes have not been studied. In this study we apply the that the periplasm and cytoplasm are iso-osmotic (31,32). method of micropipette aspiration to probe the stretching Hence the turgor pressure is supported by the peptido- elasticity of Escherichia coli (E. coli) spheroplast mem- glycan-cell wall complex, rather than by the cytoplasmic branes. The measurements reveal the basic properties of membrane (21,30). However, the physical properties of bacterial cytoplasmic membranes, which are significantly cytoplasmic membranes are otherwise unknown. different from that of red cell membranes, or of labora- Accordingly, we investigated E. coli speheroplasts over a tory-prepared lipid vesicles. It is generally believed that range of external osmolality. We found that, except at very the tension in the cytoplasmic membrane determines the low external osmolalities, the spheroplast membranes are actions of mechanosensitive channels and osmoregula- tensionless. However, increasing the area of a spheroplast tors (16–18), and perhaps other integral proteins as well membrane by micropipette aspiration creates tension. As (19,20). The cytoplasmic membranes are also the target of the tension changes by aspiration, the area of the spheroplast naturally produced membrane-acting antibiotics, as has membrane changes to a new equilibrium value by a relaxa- been demonstrated recently with LL37 (21,22), cecropin tion process. And the relaxation is loading-rate dependent. The equilibrium tension-area relation is reversible. These results indicate that the cell maintains a minimum surface Submitted May 8, 2014, and accepted for publication September 30, 2014. area without a tension, and the surface area is controlled *Correspondence: [email protected] by a membrane reservoir equivalent to membrane folds. Editor: Hagan Bayley. Ó 2014 by the Biophysical Society 0006-3495/14/11/2082/9 $2.00 http://dx.doi.org/10.1016/j.bpj.2014.09.034 E. coli Spheroplast Membrane 2083
At constant external osmolality, unfolding and refolding of water-filled U tube manometer and a negative pressure in the pipette was the membranes are reversible, thus providing an apparent produced by adjusting the height of the water level reference to the atmo- elasticity of stretching. However, when the cell lost its sphere pressure (34). Although the technique for measuring spheroplasts and giant unilamellar vesicles (GUVs) is the same, the small size of sphe- viability, the area increase loses its reversibility, implying roplasts imposes some restrictions on spheroplast measurements. For GUV that the reversible membrane reservoir is metabolically experiments, the typical dimensions are 5 mm for the micropipette radius maintained. Our results can be understood by assuming ðRpÞ and 15 mm for the GUV radius ðRvÞ. The applied membrane tension that the membrane reservoirs are mediated by reversible is in the range of 1 to 7 mN/m, which requires a suction pressure Dp of 0:27 1:9 103 2:7 18:9 noncovalent chemical bonds. Pa, equivalent to cm of water [calculated from the Laplace equation Dp ¼ 2tð1=Rp 1=RvÞ ] (see Fig. S1 A). For a spheroplast of radius 2.5 mm, using a micropipette of radius 1.1 mm, MATERIALS AND METHODS the same tension would require three times as much water pressure. Thus the maximum applicable membrane tension is limited to ~ 2 mN/m Bacterial strains and culture by the 20 cm height of the water column. If an aspirated spheroplast consisted of a spherical part and a cylindri- E. coli K-12 strain MG1655 (ATCC 700926) was purchased from ATCC cal part (a protrusion into the micropipette) (similar to the GUV shown (Manassas, VA). Luria-Bertani (LB) medium (5 g/L yeast extract, 10 g/L in Fig. S1 A), Lp the length of the protrusion, Rp the radius of the micropi- peptone from casein, and 10 g/L sodium chloride) containing 15g/L agar pette, and Rv the radius of the spherical part were carefully measured. from EMD Millipore (Billerica, MA) was used for the growth of colonies Then it was straightforward to show DA ¼ 2pRpDLp þ 8pRvDRv and 2 2 of E. coli. The medium was autoclaved before used for sterility. DV ¼ pRpDLp þ 4pRv DRv (33). As long as the osmolality balance was maintained, there should be no change of volume (the effect of the pressure Chemicals and media change by suction was so small that its contribution to the chemical poten- tial change was ~ 10 3 that of osmolality). Under the condition DV ¼ 0, DA was directly proportional to DL : DA 2pRp 1 Rp=Rv DLp. The frac- Sucrose, Tris, hydrochloric acid, lysozyme, DNase, EDTA, magnesium p ¼ ð Þ DA=A chloride, sodium hydroxide, cephalexin, carboxyfluorescein, and carbonyl tional area change was calculated from the change of the protrusion DL cyanide m-chlorophenylhydrazone (CCCP) were purchased from Sigma length p. If aspiration created a spherical protrusion (Fig. S1 B)(35), the V ph2 r h=3 pH2 R H=3 Aldrich (St. Louis, MO). FM 4-64 and Sytox green were purchased from volume was ¼ ð ð ÞÞ þ ð v ð ÞÞ, and the area A 2prh 2pR H Life Technologies (Grand Island, NY). E. coli total lipid extract was pur- ¼ þ v (Fig. S1 C); the area change was calculated at constant chased from Avanti Polar Lipids (Alabaster, AL). volume and the tension was calculated by the Laplace equation Dp ¼ 2tð1=r 1=RvÞ. The experiment of the GUVs of E. coli total lipid extract followed the Preparation of E. coli spheroplasts previously established method (34).
We prepared giant spheroplasts of E. coli by following the detailed pro- cedure described by Renner et al. (14). Briefly, cells were grown in LB me- RESULTS dium by shaking at 37 C overnight to the stationary phase. A small aliquot (1:100 dilution) was added into the LB medium and incubated in 37 C Giant spheroplasts of E. coli while shaking at 200 rpm until the absorbance at l ¼ 600 nm reached 0.5 to 0.7. One-half ml of the cell culture was diluted to 5 ml in LB medium. The methods for growing E. coli spheroplasts are well estab- m To grow long filamentous cells, cephalexin (60 g/ml) was added and the lished (14,36). First, cephalexin was added to the cell cul- culture was grown with shaking (200 rpm) at 42 C for 2 h. After cells reached an average length of ~ 50 mm, cells were harvested by centrifuga- ture. In the presence of cephalexin that blocks septation, tion at 3000 g for 1 min. The pellet was suspended in 500 mL of 800 mM E. coli grew into long filaments (Fig. 1 A). We have experi- sucrose solution. Spheroplasts from filamentous cells were formed by add- mented with the growth time and temperature (37,38)to ing the reagents in the following order: 30 mL 1 M Tris HCl (pH 8.0), achieve longer filaments (T 50 mm). To remove the outer m m m 24 L 0.5 mg/mL lysozyme, 6 L 5 mg/mL DNase, and 6 L 125 mM membrane and the peptidoglycan, we added lysozyme and EDTA-NaOH (pH 8.0). After 5 to 20 min at room temperature, 100 mL of STOP solution (10 mM Tris HCl at pH 8, 0.7 M sucrose, 20 mM EDTA to the cell suspension, that converted filaments into MgCl2) was added to stabilize the spheroplasts. We did not find noticeable spheroplasts (Fig. 1 A). Then a STOP solution (14) was differences in the results by different time duration of lysozyme treatments. added to terminate the digestion. The osmolality of the Spheroplasts were stored in liquid nitrogen for no longer than 2 weeks. STOP solution was 0.845 Osm. It has been shown previously m Frozen spheroplasts were thawed slowly on ice before each use. Thirty L that the ratio of DNA/protein in filaments was about the of spheroplasts was injected into an observation chamber containing a certain concentration of STOP solution. The spheroplasts were either directly same as in untreated cells (13). Thus ~ 30 bacterial genomes imaged with phase contrast microscopy or labeled with FM 4-64 (8 mM) were present in a long filament and also in a resulting giant for fluorescence imaging. We were able to measure the membrane area spheroplast. We confirmed that the spheroplasts could revert changes with either phase contrast or fluorescence images, although it was to normal form of E. coli when returned to growth medium easier with the latter. We did not notice any difference in the membrane prop- (13) (data not shown). We did not know if there were rem- erties of spheroplasts whether they were labeled with FM 4-64 or not. The majority of aspiration measurements were performed with FM 4-64 label. nants of outer membrane or peptidoglycan attached to the cytoplasmic membranes of spheroplasts (39). If there were, we did not detect the effect of their presence. Method of micropipette aspiration To experiment with the cytoplasmic membranes, we This method was a modification of the original method of Kwok and Evans diluted the STOP solution by adding pure water. As the (33) as described in Sun et al. (34,35). A micropipette was connected to a STOP solution was diluted, the water influx enlarged the
Biophysical Journal 107(9) 2082–2090 2084 Sun et al.
Micropipette aspiration of spheroplast membranes
The method of aspiration serves two purposes: one is to apply tension to the membrane and another is to measure the membrane area changes. Previous studies have shown that the membrane area changes by tension (40), by peptide binding (41), or by structural events (such as pore forma- tion) (42) can provide a quantitative description for the physical event taking place in the membrane. In this study we used a micropipette to apply a small negative pressure (aspiration) to a spheroplast. When a spheroplast in the 100% STOP solution was aspirated by a micropipette, the whole cell deformed and flowed into the pipette (Fig. S3). This could be because of the size of the micropi- pette being too close to the size of the spheroplast, or it could be what was described as a liquid drop model in the micropipette-aspiration experiments of human blood cells (4,5,11). We did not analyze the spheroplasts from 100% STOP solution. When the osmolality of the external medium was reduced, the cell volume visibly enlarged because of wa- ter influx. The response of the enlarged spheroplasts to micropipette aspiration was similar to the response of a GUV of lipids. In this case, aspiration reshaped the sphe- roplast surface to include a protrusion in the micropipette, leaving a larger spherical part outside the pipette (Fig. S1 A or B). The equal osmolality inside and outside should keep the volume of the spheroplast constant. Then any surface area change would result in a change in the length of the protrusion, from which we measured the fractional area change DA=A (see Materials and Methods (34,35)). FIGURE 1 E. coli spheroplasts. (A) E. coli cells, filamentous cells, sphe- We systematically investigated the spheroplast mem- roplasts in 100% STOP solution, and in diluted STOP solutions. Scale bar ¼ branes by the micropipette-aspiration method and found m 5 m. (B) Size distribution among a population in different concentrations the following general properties for E. coli spheroplast of STOP solution. As the external concentration decreased below ~ 50%, more spheroplasts appeared to have lost the interior phase contrast. The membranes. spheroplasts without interior phase contrast were not included in the radius survey. (C) The average radius of the population in different concentrations. Membrane area expansion is a relaxation process (D) Correlations of size versus phase contrast ðI0 IÞ=I0 where I was the average phase contrast intensity of cytosol, I0 that of the outside solution. We found that the initial surface tension of a spheroplast was Spheroplasts without phase contrast had ðI0 IÞ=I0(0:01. To see this always zero, except for very swollen spheroplasts that will figure in color, go online. be discussed later. When a tension was applied to the sphe- roplast surface, its surface area increased in a manner spheroplasts, indicating that there was a membrane reservoir describable as a relaxation process (Fig. 2). Depending on in the original state of spheroplast. The degree of swelling the amount of area change, it could take as long as ~ 100s was correlated to the reduction of the interior phase contrast for the surface area to reach equilibrium. If the tension of the spheroplast against the external medium. It is impor- was then reduced to the initial near zero value, the apparent tant to note that the swelling and the phase contrast changes surface area recovered the original value also in a relaxation were reversible (Fig. S2). The size distribution of the sphe- manner. This is very different from stretching a lipid bilayer roplasts is plotted for different external osmolalities in Fig. 1 (33,40), where the area change follows the tension with a B and the average size in Fig. 1 C. The correlation between speed of sound in the material (43). the size and the interior phase contrast is shown in Fig. 1 D. The relaxation process gave rise to a loading-rate depen- In general swelling increased with decreasing external dence (44,45), which is well known in single-protein pulling osmolality and swelling reduced phase contrast. However, experiments (46,47). When the tension was loaded in in a given external osmolality, there was a range of degrees different rates, the surface area expansion followed different of swelling among the spheroplast population. relaxation paths (Fig. 2).
Biophysical Journal 107(9) 2082–2090 E. coli Spheroplast Membrane 2085
FIGURE 2 Time dependent area change of spheroplast membrane under a change of tension. (A) At time 0, the tension was increased from 0.25 to 0.83 mN/m. The images from top to bottom were at t ¼ 0, 5, and 70 s. Scale bar ¼ 2.5 mm. (B) After the membrane area reached equilibrium, the tension was decreased to 0.25 mN/m. The images from top to bottom were at t ¼ 0, 5, and 70 s. The dash-dot lines indicate the geometric features used to calcu- late the area changes. (C) Two area versus time curves were performed at different rates of applying tension: red, the tension (from 0.25 to 0.83 mN/m) was applied by lowing the water column at 2 mm/s; blue, at 10 mm/s (see Materials and Methods). The curve fitting used the formula described in Discussion; the time constants for the blue curve are 14.6 and 8.4 s for area increase and decrease, respectively. 28.8 and 20.9 s for the red curve, respectively. To see this figure in color, go online.
Surface area expansion by tension is reversible Every spheroplast in a constant external osmolality gave a well-defined equilibrium tension t versus DA=A curve. The curve is approximately linear for a small range of t. (The range of t was limited by the height of the manometer water column used to generate the aspiration pressure—see FIGURE 3 (A) Three examples of reversible tension t versus fractional Materials and Methods). Importantly we found that the area change DA=A of spheroplast membranes. In each example blue data stress-strain relation is reversible and repeatable (Figs. 3 are for increasing tension and red for decreasing tension. The data points and S4) within errors. From the linear relation, we define are equilibrium values. The error bars represent the uncertainties in calcu- lating the area from the microscope images. The uncertainties in tension the apparent stretching modulus Ka ¼ t=ðDA=AÞ.We calculation are smaller than the data symbols. Ka, the apparent stretching measured Ka of the spheroplasts taken from a population modulus, was measured from the slope of each curve. (B) Ka of a sphero- equilibrated in external osmolality 0.237 Osm (Fig. 3 B). plast population in 28% STOP solution. To see this figure in color, go online. The largest values of Ka from the population are ~ 100 mN/m. In comparison, the typical values of Ka for lipid bilayers are ~ 200 mN/m (40). Only very swollen sphero- plasts. The requirement of a 100 microscope objective plasts exhibit Ka values in the same range (see below). and its short working distance made it very difficult to trans- fer an aspirated spheroplast from one chamber to another, as Temperature dependence we did with GUVs (34). Also, an aspirated spheroplast was Bacterial membranes may have a gel-fluid phase transition easily detached from the micropipette by a flow in the cham- below the growth temperature (48). Thus it is important to ber, making it difficult to change the solution in the sample test the possible temperature dependence of their elastic prop- chamber. We could only change the solution osmolality erties. A number of randomly chosen spheroplasts were qualitatively. This was done by introducing an open tube measured by t versus DA=A, first in room temperature 25 C containing a 100% STOP solution that included 1 mM car- then at 37 C, the growth temperature (Fig. 4). In all cases, boxyfluorescein (CF) into the chamber. We let the diffusion we detected no significant temperature dependence. Thus from the tube slowly increased the osmolality inside the we performed the rest of experiments in room temperature. chamber until the fluorescence intensity of CF in the cham- ber appeared to be uniform. We measured Ka of a sphero- Stretching modulus Ka changes with osmolality plast before and after the introduction of the open tube. Measurements of the same spheroplast in different osmolal- We found that Ka invariably decreased if the external osmo- ities were difficult because of the small size of the sphero- lality was increased (Fig. 5).
Biophysical Journal 107(9) 2082–2090 2086 Sun et al.